286 High Performance Visualization
munication can take place without any operating system or MPI overhead like
message buffering and so forth. A second common trait concerns the use of
ghost zones. With a data decomposition consisting of smaller and more data
partitions, there is more surface area compared to a decomposition that re-
sults in fewer and larger data partitions. Less surface area means there is less
information—ghost zones—that needs to be communicated during the course
of processing. The MPI-only configuration results in more and smaller data
partitions compared to the MPI-hybrid configurations.
In the future, all trends suggest computational platforms comprised of an
increasing number of cores per chip (see Chap. 15). One unknown is whether or
not those future architectures will continue to support a shared memory that is
visible to all cores. The present hybrid-parallel implementations perform well
because all threads have access to a single shared memory on a CPU chip. If
future architectures eliminate this shared memory, future research will need to
explore alternative algorithmic formulations and implementations that both
exploit available architectural traits as well as achieve low memory footprint
utilization and reduced communication when compared to traditional, MPI-
only designs and implementations.
Hybrid Parallelism 287
References
[1] VisIt – Software that delivers Parallel, Interactive Visualization.
http://visit.llnl.gov/.
[2] C. Bajaj, I. Ihm, G. Joo, and S. Park. Parallel Ray Casting of Visi-
bly Human on Distributed Memory Architectures. In VisSym ’99 Joint
EUROGRAPHICS-IEEE TVCG Symposium on Visualization, pages
269–276, 1999.
[3] OpenMP Architecture Review Board. OpenMP Application Pro-
gram Interface Version 3.1, July 2011. http://www.openmp.org/wp/
openmp-specifications.
[4] David R. Butenhof. Programming with POSIX threads. Addison-Wesley
Longman Publishing Co., Inc., Boston, MA, USA, 1997.
[5] David Camp, Hank Childs, Amit Chourasia, Christoph Garth, and Ken-
neth I. Joy. Evaluating the Benefits of an Extended Memory Hierarchy for
Parallel Streamline Algorithms. In Proceedings of the IEEE Symposium
on Large-Scale Data Analysis and Visualization (LDAV), Providence, RI,
USA, October 2011.
[6] David Camp, Christoph Garth, Hank Childs, Dave Pugmire, and Ken Joy.
Streamline Integration Using MPI-Hybrid Parallelism on a Large Multi-
Core Architecture. IEEE Transactions on Visualization and Computer
Graphics, 17(11):1702–1713, November 2011.
[7] Robit Chandra, Leonardo Dagum, Dave Kohr, Dror Maydan, Jeff Mc-
Donald, and Ramesh Menon. Parallel Programming in OpenMP. Morgan
Kaufmann Publishers Inc., San Francisco, CA, USA, 2001.
[8] Hank Childs, Eric S. Brugger, Kathleen S. Bonnell, Jeremy S. Meredith,
Mark Miller, Brad J. Whitlock, and Nelson Max. A Contract-Based Sys-
tem for Large Data Visualization. In Proceedings of IEEE Visualization,
pages 190–198, 2005.
[9] NVIDIA Corporation. What is CUDA? http://www.nvidia.com/
object/what_is_cuda_new.html, 2011.
[10] Robert A. Drebin, Loren Carpenter, and Pat Hanrahan. Volume render-
ing. SIGGRAPH Computer Graphics, 22(4):65–74, 1988.
[11] P. Fischer, J. Lottes, D. Pointer, and A. Siegel. Petascale Algorithms for
Reactor Hydrodynamics. Journal of Physics: Conference Series, 125:1–5,
2008.
288 High Performance Visualization
[12] Khronos Group. OpenCL – The Open Standard for Parallel Programming
of Heterogeneous Systems. http://www.khronos.org/opencl/, 2011.
[13] Mark Howison, E. Wes Bethel, and Hank Childs. MPI-hybrid Paral-
lelism for Volume Rendering on Large, Multi-core Systems. In Euro-
graphics Symposium on Parallel Graphics and Visualization (EGPGV),
Norrk¨oping, Sweden, May 2010. LBNL-3297E.
[14] Mark Howison, E. Wes Bethel, and Hank Childs. Hybrid Parallelism
for Volume Rendering on Large, Multi- and Many-core Systems. IEEE
Transactions on Visualization and Computer Graphics, 99(PrePrints),
2011.
[15] Marc Levoy. Display of Surfaces from Volume Data. IEEE Computer
Graphics and Applications, 8(3):29–37, May 1988.
[16] Kwan-Liu Ma. Parallel Volume Ray-Casting for Unstructured-Grid Data
on Distributed-Memory Architectures. In PRS ’95: Proceedings of the
IEEE Symposium on Parallel Rendering, pages 23–30, New York, NY,
USA, 1995. ACM.
[17] Kwan-Liu Ma, James S. Painter, Charles D. Hansen, and Michael F.
Krogh. A Data Distributed, Parallel Algorithm for Ray-Traced Volume
Rendering. In Proceedings of the 1993 Parallel Rendering Symposium,
pages 15–22. ACM Press, October 1993.
[18] Jason Nieh and Marc Levoy. Volume Rendering on Scalable Shared-
Memory MIMD Architectures. In Proceedings of the 1992 Workshop on
Volume Visualization, pages 17–24. ACM SIGGRAPH, October 1992.
[19] NVIDIA Corporation. NVIDIA CUDA
TM
Programming Guide Ver-
sion 3.0, 2010. http://developer.nvidia.com/object/cuda_3_0_
downloads.html.
[20] Thomas Porter and Tom Duff. Compositing Digital Images. Computer
Graphics, 18(3):253–259, 1984. Proceedings of ACM/Siggraph.
[21] Dave Pugmire, Hank Childs, Christoph Garth, Sean Ahern, and Gun-
ther H. Weber. Scalable Computation of Streamlines on Very Large
Datasets. In Proceedings of Supercomputing (SC09), Portland, OR, USA,
November 2009.
[22] Paolo Sabella. A Rendering Algorithm for Visualizing 3D Scalar Fields.
SIGGRAPH Computer Graphics, 22(4):51–58, 1988.
[23] Marc Snir, Steve Otto, Steven Huss-Lederman, David Walker, and Jack
Dongarra. MPI – The Complete Reference: The MPI Core, 2nd ed. MIT
Press, Cambridge, MA, USA, 1998.
Hybrid Parallelism 289
[24] R. Tiwari and T. L. Huntsberger. A Distributed Memory Algorithm for
Volume Rendering. In Scalable High Performance Computing Conference,
Knoxville, TN, USA, May 1994.
[25] The Top 500 Supercomputers, 2011. http://www.top500.org.
[26] Craig Upson and Michael Keeler. V-buffer: Visible Volume Rendering.
In SIGGRAPH ’88: Proceedings of the 15th Annual Conference on Com-
puter Graphics and Interactive Techniques, pages 59–64, New York, NY,
USA, 1988. ACM.
..................Content has been hidden....................
You can't read the all page of ebook, please click here login for view all page.